Ecology

10 SPECIAL FEATURE Vol. 79, No. 1

Ecology, 79(1), 1998, pp. 10–22

q 1998 by the Ecological Society of America

EXPERIMENTAL INVESTIGATION OF NUTRIENT LIMITATION OF

FOREST GROWTH ON WET TROPICAL MOUNTAINS
E. V. J. TANNER,1 P. M. VITOUSEK,2 AND E. CUEVAS3
1Department of Plant Sciences, University of Cambridge, Downing Street, Cambridge CB2 3EA, UK
2Department of Biological Sciences, Stanford University, Stanford, California 94305-5020 USA
3Centro de Ecologia, Instituto Venezolano de Investigaciones Cientificas, Apartado 1827,

Caracas 1010-A, Venezuela

Abstract. This article reviews literature and summarizes experiments to investigate the
extent to which productivity of tropical montane rain forests is constrained by low nutrient
supply. On any one mountain, with increase in altitude foliar N decreases, and P and K
usually decrease, but Ca and Mg show no consistent trend. However for a wide range of
sites N, P, K, Mg, and Ca show no trends. Litterfall contents of N and P and often K, Ca,
and Mg are lower in montane forests than in lowland forests, mainly because of reduced
litterfall mass, but N and P concentrations are also lower in forests above 1500 m. Tropical
montane soils usually have more soil organic matter per unit ground area; N mineralization
levels are lower at higher altitudes in Costa Rica, and extractable and total soil P are lower
in sites with lower litterfall P concentrations. We speculate that many lowland forests are
limited by P and many montane forests by N.
Fertilization studies on ash-derived montane soils in Hawai‘i showed a trend for a switch
from N limitation on young soils to P, or N and P, limitation on soils over older substrates.
Jamaican montane trees were limited by N and by P separately. Venezuelan montane trees
were limited by N. The sites in Jamaica and Venezuela have soils of indeterminate age.
Taken together these results show that nutrient limitation is widespread in montane soils
(all sites have responded to at least one nutrient) and that the particular nutrient(s) that
limit(s) production may differ for explicable reasons. First results from lowland forests on
sandy soils in Kalimantan show N or simultaneous N and P limitation. Many more exper-
iments, especially in lowland forests, are needed to test our speculation that P usually limits
productivity in tropical lowland rain forests and that N limits productivity in tropical
montane rain forests.
Key words: fertilization experiments; forests in Hawai‘i, Jamaica, and Venezuela; growth limi-
tation; montane cf. lowland forests; nitrogen; nutrient limitation of forest growth; nutrients; phos-
phorus; soil age and nutrient limitation; tropical montane rain forest productivity.

INTRODUCTION studies because of their importance in the determina-

Among the many factors that vary from lowland to
montane tropical forests, decreasing air temperature
and increasing cloudiness are probably ultimately—
and may be primarily—responsible for limiting the
growth and distribution of tropical montane forests
(Grubb 1977). Nutrient supply and other factors such
as soil saturation are also controlled by temperature
and precipitation and may be simultaneously and even
proximally limiting. This article investigates the extent
to which the productivity of montane tropical forest is
constrained by low nutrient availability. We briefly re-
view observational evidence and concentrate on recent
experimental studies where forests have been fertilized.
We attach particular importance to these experimental
Manuscript received 15 November 1995; revised 9 January
1997; accepted 14 February 1997; final version received 5
May 1997. For reprints of this Special Feature, see footnote
1, page 1.
10
tion of causation.
Limitation by a nutrient is shown if the rate of an
ecosystem process is increased by addition of that nu-
trient, and strictly speaking it can only be determined
experimentally. Importantly, limitation can be seen in
the response of organisms that already occupy a site,
or only after colonization by organisms that are better
able to make use of the added resource (Chapin et al.
1986, Berendse 1993). Furthermore while overall eco-
system processes may be nutrient limited, not all spe-
cies in the ecosystem need be limited; indeed even
within a species some individuals could be limited and
others not, due, for example, to different crown ex-
posure. Finally, the observation that a nutrient is lim-
iting does not mean that only that nutrient limits the
ecosystem—simultaneous limitation by multiple re-
sources is the rule (Bloom et. al. 1980, Field et al.
1992), especially with different types of resources (e.g.,
 January 1998 TROPICAL MONTANE FORESTS
photosynthetically active radiation, water, depletable
resources like N or P, and non-depletable resources like
CO2).
It will always be easier to assess the nutrient status
of sites or individual trees by measurements, such as
foliar and litter concentrations or soil chemical ex-
tracts, rather than by experimentation. In general, nu-
trient limitation of growth is correlated with low con-
centrations of the limiting nutrient in leaves and leaf
litter (and often wood and roots); the low litter con-
centrations are a result of low leaf concentrations and
sometimes (but not always) higher rates of retranslo-
cation of nutrients from leaves during senescence.
There are important exceptions to the generalization
that nutrient limitation should result in low concentra-
tions of the limiting nutrient in plant parts. This is
especially true for N; extra N can cause extra growth
but no change in N concentration (e.g., Healey 1989).
Soil chemical extracts are more difficult to interpret
than leaf and litter concentrations, but sometimes low
levels of nutrients extracted with mild extractants (wa-
ter, dilute salt solutions) are indicative of nutrient lim-
itation. For soils, comparability between studies is lim-
ited by the many different extractants used for the same
nutrient. Despite these limitations it is important to
look for correlations between leaf, litter, and soil nu-
trients on the one hand and fertilizer responses on the
other. Where good correlations exist they are useful for
predicting which nutrients limit and to what extent, in
areas where fertilizer experiments have not been car-
ried out.
CHANGES INFOREST STRUCTURE ON ASCENDING
WET TROPICAL MOUNTAINS
The average trend in the wet tropics is for forests to
decrease in stature, growth, and productivity with in-
crease in altitude, but there is much variation at any
given altitude due to differences in nutrient availability,
water availability, and species composition. In the low-
lands the forests range in canopy height (excluding
emergents) from a few meters (Bana forest in southern
Venezuela) to 40 m tall (in many parts of lowland
Southeast Asia). In the mountains, up to 2500 m in
altitude, forests can be up to 30–35 m tall. Only above
3000 m are forests with canopy heights .30 m absent;
even at 3370 m there is a forest with a canopy height
of 22 m (in Colombia; Veneklaas 1991).
FOLIAR NUTRIENTS
On any one mountain system with increase in altitude
there is usually a decrease in concentrations (on a mass
basis) of foliar N (5 of 5 transects in Fig. 1); P and K
often fall (3 of 5 transects); but there are no consistent
trends for Ca and Mg. However when data from a wide
range of sites are combined, foliar nutrient concentra-
tions show no relationship with altitude for N (as found
for evergreen woody plants by Körner [1989]), P, K,
11
Mg, and Ca (Fig. 2). The differences between specific
transects and the general pattern shows that one has to
be especially careful to make relevant comparisons for
N and P (usually specific transects). At any altitude it
is possible to find forests with low, intermediate, and
high concentrations of nutrients, but at any altitude
low-stature forests, especially those where canopy
height is ,10 m, have low concentrations of N and P
(Fig. 3); they also have low concentrations of K, Ca,
and Mg (but for these three nutrients there are some
tall forests that have low concentrations of these nu-
trients).
In some ways it makes more sense to express nutrient
concentrations per leaf area, because the primary func-
tion of leaves is to absorb light, which is mostly a result
of area and only slightly increased by increasing leaf
thickness. Unfortunately the necessary measurement of
leaf mass per area is only occasionally made. Further-
more leaf mass per area varies so much through the
canopy from sun to shade leaves that it is essential to
measure leaf mass per area on the sample being ana-
lyzed for nutrients, and to collect fully sunlit leaves.
Nutrient concentrations per mass show little difference
between sun and shade leaves for N, P, S, Fe, and Cu,
while shade leaves have higher concentrations of K,
Mg, and Mo (Grubb 1977).
At least two papers present nutrient concentrations
from tropical montane systems by leaf area and mass.
In Hawai‘i on a mass basis N and P concentrations
decreased with altitude, but on an area basis concen-
trations increased with altitude particularly above 1600
m (Vitousek et al. 1992). In Puerto Rico, on a mass
basis N decreased, and P, K, Ca, and Mg did not change
with increase in altitude, but on an area basis all five
elements increased with altitude (Medina et al. 1981).
In both places leaf mass per area increased with in-
crease in altitude. The higher leaf mass per area is in
turn correlated with longer-lived leaves in tropical
montane forests; for example in ‘Mor Ridge forest’ (at
1550 m in Jamaica) leaves last for 18 mo on the average
(Tanner 1980)—much longer than the average leaf life
in tropical lowland rain forests. It seems likely from
these results that the instantaneous nutrient-use effi-
ciency (the rate of CO2 fixation per unit nutrient) de-
creases with increase in altitude; this is partly due to
lower temperatures and partly due to the short-term
inefficiency of sclerophyllous leaves.
PRODUCTIVITY, NUTRIENT CYCLING, AND NUTRIENT
RETRANSLOCATION
Tropical montane forests generally have reduced pro-
ductivity compared to lowland forests (Tanner 1980).
The major components of net primary production
(NPP) in tropical montane forests are, in order, above-
ground litterfall, root production, and, to a much lesser
extent, wood increment; wood increment is particularly
low in montane forests compared to lowland forests.
 Ecology
12 SPECIAL FEATURE Vol. 79, No. 1

FIG. 1. Foliar nutrient concentrations and altitude in

forests on transects on the same mountain. Data for Silam
are from Proctor et al. (1989), for Hawai‘i from Vitousek
et al. (1992), for Puerto Rico from Medina et al. (1981),
and for Malaya and New Britain/New Guinea from Grubb
(1977).
January 1998 TROPICAL MONTANE FORESTS 13

FIG. 2. Foliar nutrient concentrations and altitude in

forests on various mountains in the wet tropics (usually
not transects on the same mountain). Data are from Grubb
(1977: Table 4). In the leaf N panel additional sources of
data are: 15 Veneklaas (1991), 2 5 Bruijnzeel and Proctor
(1993), 3 5 Nadkarni and Matelson (1992), 4 5 Vitousek
et al. (1995b), 5 5 Tanner (1977), 6 5 Proctor et al.
(1989), 7 5 70 km north of Manaus, Brazil (M. Davies
and E. V. J. Tanner, unpublished data), 8 5 Grubb et al.
(1994), 9 5 near Mérida, Venezuela (E. V. J. Tanner, un-
published data).
 Ecology
14 SPECIAL FEATURE Vol. 79, No. 1

FIG. 3. Foliar nutrient concentrations and canopy

heights (excluding emergents) in forests at various alti-
tudes in the wet tropics. Data sources are as in Fig. 2.
 January 1998 TROPICAL MONTANE FORESTS 15

altitudes; Fig. 4). Transects on the same mountain show

lower concentrations of N, but usually not of P, in litter
at higher altitudes (Fig. 5). Note that the patterns for
P concentrations in litter, lower at the highest altitudes
(wide range of sites) or no change with altitude (on the
same mountain) contrast with those for foliar P con-
centrations, which show no change with altitude (wide
range of sites) or fall with altitude (on the same moun-
tain). These results conflict because the data for single
mountains suggest less retranslocation of P prior to
abscission at higher altitude, whereas the pattern from
a wide range of sites suggest more retranslocation of
P at high altitude. Clearly more data are needed es-
pecially from transects on the same mountain (note that
for P only one site, Silam, is in both Figs. 1 and 5).
In a steady state the amount of a nutrient in litterfall
(per unit ground area per time) is a reasonable ap-
proximation to net nutrient uptake into aboveground
parts (per unit ground area per time). Based on litterfall,

FIG. 4. Litterfall nutrient contents and concentrations in

a large number of wet tropical forests, showing that for both
N and P higher contents are correlated with higher concen-
trations, and that forests from the highest altitudes (.1500
m) have low contents and concentrations for both N and P.
Data are from Vitousek (1984: Table 2; the present graph has
most of the same data as Vitousek’s Fig. 1; the present graph
plots mg/g on the y axis whereas Vitousek’s graph plots the
reciprocal of mg/g [5dry mass per unit nutrient]).

Our information on the components of NPP is best for

litterfall, which has been widely measured using rea-
sonably comparable techniques. Caution is required
when generalizing from only one of the major com-
ponents of NPP, but the patterns are striking. Montane
forests generally have reduced cycling of N and P and
often of K, Ca, and Mg, in comparison with lowland
forests (Grubb 1977, Vitousek 1984). The main reason
for this is reduced rates of litterfall, but N and P con-
FIG. 5. Litterfall nutrient concentrations and altitude in
centrations in litter are lower in forests from the highest tropical forests on transects on the same mountain. Redrawn
(.1500 m) altitudes (though concentrations are not from Bruijnzeel and Proctor (1993); Mulu data are from Proc-
systematically lower at intermediate [1000–1500 m] tor et al. (1983).

16 SPECIAL FEATURE
N and P and sometimes K, Ca, and Mg uptake is lower
in montane forests (for transects on individual moun-
tains [Fig. 6]; though for a wide range of sites on many
different mountains there is no general trend except for
low values of N and P above 1500 m [Fig 4]). For N
and P, and sometimes K, Ca, and Mg, these low
amounts result in low concentrations in the leaves (but
there is a problem of a chicken and egg argument here).
A HYPOTHESIS THAT IN GENERAL P LIMITS
GROWTH IN LOWLAND FORESTS AND
N IN MONTANE FORESTS
A reasonable speculation from the patterns of foliar
and litterfall concentrations is that N is in excess supply
in most lowland rain forests, a speculation supported
by the relatively high rates of N mineralization and
N-containing trace gas emissions from the soils (a by-
product of high rates of nitrogen transformations) in
such sites (Matson and Vitousek 1987). In contrast,
many tropical montane forests seem to be likely to
suffer from insufficient N. A further speculation is that
P is likely to limit growth in a substantial subset of
lowland rainforests, and N in many montane forests.
Thus, with regard to N, tropical montane forests ap-
pear to function more like many temperate and boreal
forests than like most lowland tropical forests. Tropical
lowland rain forests produce more litter annually than
do most temperate forests—and in most cases they pro-
duce litter with much higher concentrations of N (lower
C:N ratios) than do temperate or boreal forests. Phos-
phorus and other major elements (Ca, Mg, K) are more
variable—many tropical forests produce litter with
much lower concentrations of P than most temperate
forests but others produce litter with higher concen-
trations (Vitousek 1982). This variation among tropical
lowland rain forests is associated with soil type and
fertility; tropical forests on highly weathered clay soils
(oxisols and ultisols) generally produce litter low in P,
those on richer soils (i.e., alfisols) produce litter with
high P concentrations, and those on sandy soils are
often low in N (Vitousek and Sanford 1986).
TROPICAL MONTANE SOILS
The decreased nutrient concentrations in leaves and
decreased cycling of nutrients in litter that occur as one
ascends tropical mountains could be due to immobi-
lization in the soil, which in turn is due to a buildup
of soil organic matter resulting from incomplete de-
composition. Montane soils usually have more soil or-
ganic matter per unit ground area (e.g., in Malaya,
Whitmore and Burnham 1969; and in Thailand, Kira
and Shidei 1967); their organic matter is also often
concentrated near the top of, or above, the mineral
layers of the soil. Another reason for supposing that
nutrient supplies might be limited in montane soils is
that many montane soils are of low pH. (In such soils
the low pH is a result of the leaching of bases and/or
Ecology
Vol. 79, No. 1
the high cation-exchange capacity due to the high or-
ganic-matter concentration; low pH is not the cause of
nutrient limitation—low pH is caused by the same fac-
tors that cause nutrient limitation; pH is easily mea-
sured and may indicate nutrient shortage). Often there
is a decrease in nutrient status as one moves from sites
with higher to lower pH, especially when pHs are below
5. For example in three ridge-top sites in close prox-
imity in Jamaica the surface soil pHH2O (i.e., pH mea-
sured in water) falls from 4.3 to 3.0 and average foliar
N falls from 1.76% to 1.05% and average foliar P from
0.10% to 0.05%; the most stunted forest, Mor Ridge
forest, has soil with the lowest pH (Tanner 1977). How-
ever, not all low-pH soils have stunted forests; e.g.,
both the extremely stunted forest on East Peak at 1034
m in Puerto Rico (Lyford 1969) and the much better
developed El Verde forest at 510 m on the same moun-
tain (Edmisten 1970) have the same surface soil pH
(4.3).
If immobilization in soils per se is the cause of low
availability of N, P, K, and Ca, why have trees been
unable to accumulate these elements slowly, over the
centuries, from weathering rock fragments, which are
often seen near the surface in montane forests, and from
inputs in rainfall and/or dust, and from N fixation? This
question is not specific to montane forests; vegetation
throughout the world growing on soils developed from
infertile substrates can be shown to be nutrient limited
despite millenia of rainfall, dust inputs, and N fixation.
The extra nutrients could be insufficiently available to
plants because of their form or because they are at
concentrations so low as to be unavailable, or because
the rooting medium somehow limits root uptake. In-
frequent but severe disturbance could also lead to
losses of accumulated nutrients. Alternatively Hedin et
al. (1995) recently demonstrated that old-growth Chil-
ean forests lose dissolved organic N in streams at about
the same rate that they receive (mostly inorganic) N in
precipitation; they speculate that losses of dissolved
organic nitrogen (and dissolved organic phosphorus)
could be sufficient to maintain nutrient limitation in
the long term, at least in unpolluted regions.
Data on levels of instantaneously extractable soil
nutrients in montane tropical forests are few, and some-
times show inconsistent patterns with altitude (Bru-
ijnzeel and Proctor 1993). Measures of N mineraliza-
tion made in Costa Rica showed much reduced rates
at higher altitude (Marrs et al. 1988), which correlated
with reduced amounts of N cycling in litterfall at higher
altitude on the same transect. Extractable and total soil
P were lower in sites with higher P nutrient-use effi-
ciency (litterfall dry mass divided by litterfall P
amount) both for all tropical rain forest sites and for
tropical montane rain forest sites (Silver 1994), which
suggests that P may limit growth in many lowland and
some montane sites. Exchangeable basic metals have
been found to both increase and decrease with in-
January 1998 TROPICAL MONTANE FORESTS 17

FIG. 6. Litterfall nutrient contents and altitude in trop-

ical forests on transects on the same mountain. Silam data
are from Proctor et al. (1989), Barva data from Heaney
and Proctor (1989), and Colombia data from Veneklaas
(1991).

18 SPECIAL FEATURE
creased altitude on different mountains (Bruijnzeel and
Proctor 1993). However, there is little evidence that K,
Ca, or Mg limit growth or affect composition in natural
(as opposed to plantation) forests in the tropics—a no-
table exception being the root-ingrowth experiments of
Cuevas and Medina (1988), which showed that roots
in Tierra Firme forest in southern Venezuela grew pref-
erentially into cylinders of vermiculite doped with Ca
(or potassium dihydrogen phosphate). The interpreta-
tion of this experiment was considerably strengthened
by results from Hawai‘i where root growth into bags
doped with an element was correlated with increased
growth caused by fertilization with that element (Raich
et al. 1994). Thus with a few exceptions the data on
extractable soil nutrients do not add much to the in-
terpretation based on foliar and litterfall nutrients.
Where they are available, soil nutrient data generally
support the suggestion that tropical montane rain-forest
soils have less available N than many tropical lowland
rain-forest soils though there are some montane sites
where P is likely to limit growth as well as N.
Bruijnzeel et al. (1993) have suggested that the high
phenolic concentrations in many montane soils might
slow decomposition and plant growth, but there is no
experimental evidence for this as yet. Much earlier Jor-
dan et al. (1979) showed that nitrification rates in Am-
azonian rain forest were suppressed, and they hypoth-
esized that this was a result of high tannin concentra-
tions in the root mat; they proposed that this limitation
of nitrification was a way of limiting nitrogen losses
by limiting the production of highly mobile nitrate ions.
The suggestion that low vapor-pressure deficit (the
driving force for transpiration) might limit nutrient
movement into leaves due to reduced transpiration was
shown to be incorrect by Medina et al. (1981) who
measured significant transpiration in the exceptionally
stunted forests, which are very frequently clouded, on
Pico del Este in Puerto Rico. This is the same conclu-
sion as that reached by Grubb (1977) who reported
various experiments and observations that showed that
transpiration rates were not directly related to the
amounts of ions transported, and by Smith (1991) who
reviewed the relationship between ion transport and the
transpiration stream and concluded that as the rate of
flow decreased the concentration in the xylem sap in-
creased, thus maintaining a supply of nutrients to even
slowly transpiring leaves. Thus it seems very unlikely
that low rates of transpiration might limit nutrient sup-
plies to leaves especially because the highest require-
ments will be when leaves are developing—when they
have very low rates of transpiration because they do
not have fully functional stomata and because the air
inside buds is near saturation and thus there is almost
no driving force for transpiration.
FERTILIZATION EXPERIMENTS
Experiments are the only way in which these and
other possible causes of decreased productivity and nu-
Ecology
Vol. 79, No. 1
trient cycling in montane tropical forests can be ad-
dressed directly. Nutrients are the major resource most
amenable to experimental manipulation, and a number
of fertilization experiments have been carried out in
montane tropical forests. However, the design and in-
terpretation of forest fertilization experiments is not
always straightforward, for several reasons.
1) Trees adapted to infertile soils may respond less
to nutrient enrichment than those adapted to more fer-
tile soils. Consequently, fertilization of the most nu-
trient-deficient sites may yield only a modest re-
sponse—until the dominant species are replaced by
others that are capable of responding to (and may in-
deed require) higher levels of nutrients (Chapin et al.
1986). The evidence for the importance of this pattern
is mixed—for example, Pinus nigra will grow on very
infertile soils derived from blown sand in Scotland but
it responds strongly to fertilization (Miller et al. 1976);
how to interpret these results is unclear, however, be-
cause the trees may have come from natural popula-
tions growing on relatively fertile soils.
In the experiments that we discuss below there was
often a lag time in response; this was due to the slow
response of the individuals present when the plots were
first fertilized, not to the arrival of species or to changes
in species relative abundance.
2) Phosphorus-fixing soils are widespread in tropi-
cal regions (Uehara and Gillman 1981, Sollins et al.
1988), and even relatively large additions of this ele-
ment may not increase plant-available P significantly.
3) It is difficult to measure belowground production
in forests—particularly in fertilized plots, in which cal-
culations based on CO2 emissions (cf. Raich and Na-
delhoffer 1989) are inappropriate because increased nu-
trient availability may affect soil carbon pools with
long turnover times. Increases in response to fertiliza-
tion in aboveground production alone may simply re-
flect reallocation of carbon from belowground.
4) The diversity of tree species in many tropical for-
ests makes it difficult to track species-level responses
across treatment plots of any practical size. The pos-
sibility that different species in a site might respond to
different nutrients further makes it difficult to deter-
mine ecosystem-level responses.
Despite the potential problems with the interpreta-
tion of fertilization studies, cautious insights can be
drawn from several that have been performed in trop-
ical montane forests, mostly in the American tropics
or Hawai‘i. In Jamaican montane forest with average
foliar concentrations of N, P, Ca, and Mg and high
concentrations of K (compared to the range of values
for tropical forests in Fig. 2), and low litter concen-
trations for N and especially so for P, trunk growth was
77% more following N fertilization, and increased by
32% following P fertilization (Table 1). In Venezuelan
montane forest with average foliar concentrations of N
and P, and near-average litterfall concentrations of N
 January 1998 TROPICAL MONTANE FORESTS 19

TABLE 1. Responses of various natural tropical rain forests to fertilization. Percentage increases in trunk growth and litterfall
were calculated from results from control and fertilized plots (or individuals—some experiments used individuals not
plots). Significance levels were often calculated for overall N or P effects (which are more powerful), not simply pairwise
comparison; sig. 5 significant at 0.05, NS 5 not significant at 0.05. For litterfall in Venezuela, litterfall in year 1 was
standardized to that in control plots, and litterfall for year 4 was multiplied by that standardization factor to eliminate
differences in litterfall between treatments in year 1 (before fertilization had a significant effect); this procedure was also
adopted for litterfall in the Hawaiian data from Vitousek et al. (1995a).

Increase in N in P in
trunk-diameter Increase in leaves leaves
Site Data source Soil age (yr) Substrate Fertilizer growth (%) litterfall (%) (%) (%)
Hawai‘i Puu Gerrish et al. 1500–3000 Lava Comp.† 167 (sig.) ··· 0.61 0.08
Makaala 1988
Hawai‘i Saddle Gerrish et al. 3400 Lava Comp.† 25 (NS) ··· 0.78 0.07
Road 1988
Hawai‘i Thurston Gerrish et al. 200 Ash Comp.† 87 (sig.) ··· 0.73 0.042
lava tube 1988
Hawai‘i Thurston Vitousek et al. 200 Ash N 140 (overall N 15 (overall N ··· ···
lava tube 1993 effect sig.) effect sig.)
P 20 (overall P ef- 217 (overall P ··· ···
fect NS) effect NS)
N and P 180 (interaction 29 (interaction ··· ···
NS) NS)
Hawai‘i Mauna Raich et al. 140 ’a’a lava N 357 (overall N ··· 0.62 0.068
Loa 1996 effect sig.)
P 100 (overall P ··· ··· ···
effect NS)
N and P 671 (interaction ··· ··· ···
NS)
140 Pahoehoe N 260 (overall N ··· 0.71 0.073
lava effect sig.)
P 160 (overall P ··· ··· ···
effect sig.)
N and P 680 (interaction ··· ··· ···
sig.)
Hawai‘i Mauna Vitousek and 20 000 Ash N 75 (overall N ··· 1.17 0.09
Kea Farrington effect sig.)
1997
P 13 (overall P ef- ··· ··· ···
fect NS)
N and P 188 (interaction ··· ··· ···
NS)
Hawai‘i Olaa Vitousek et al. 2000 Ash N 25 (NS) 24 (NS) 1.1 0.088
1987, Vitou-
sek et al.
1995a
Hawai‘i Kauai Herbert and 4.1 3 106 ··· N 4 (overall N ef- 23 (overall N 0.80 0.058
Fownes 1995 fect sig.) effect sig.)
P 232 (overall P 22 (overall P ··· ···
effect sig. and effect sig.)
positive)
N and P 24 48 ··· ···
Venezuela Tanner et al. ··· ··· N ··· 35 (overall N 1.6 0.081
1992 effect sig.)
P ··· 23 (overall P ··· ···
effect NS)
N and P 133 (sig.) 42 (interaction ··· ···
NS)
Jamaica Tanner et al. ··· ··· N 77 (sig.) ··· 1.6 0.07
1990
P 32 (sig.) ··· ··· ···
† Comp. 5 compound; containing N, P, K, Ca, S, Cu, Mg, Zn, and B.

and P, trunk growth was increased by 133% by simul-
taneous fertilization with N and P (Table 1), and lit-
terfall was increased by 35% by fertilization with N,
and by 42% by fertilization with N and P (Table 1).
In Hawai‘i there have been several fertilization ex-
periments in forests dominated by one species, Me-
trosideros polymorpha, which were designed to inves-
tigate (a) limitation of growth on soils of different ages
and (b) limitation of growth on soils derived from phys-
ically different (but chemically similar) substrates.

20 SPECIAL FEATURE
On soils of different ages (and different substrates)
Gerrish et al. (1988), using compound fertilizers,
showed that tree-diameter growth was limited on both
200-yr-old soil derived from ash, and 1500–3000 yr
old soils derived from lava, though it was not on a
3400-yr-old soil (also derived from lava but the trees
at that site had higher concentrations of N in their
leaves, Table 1). In later experiments with individual
nutrients and a larger range of substrate ages (on ash-
derived soils) Vitousek and co-workers (Table 1)
showed a trend for a switch from N limitation on young
soils to P, or N and P, limitation on old soils. On 30-
and 200-yr-old ash both trunk growth and litterfall were
increased by fertilization with N; on 2000-yr-old ash
N fertilization had no effect on either trunk growth or
litterfall; on 4.1 million yr old substrate P increased
trunk growth and P and N increased litterfall. These
responses to fertilization are consistent with N con-
centrations in leaves being lower in the youngest site,
but not with P concentrations in leaves, which were
also lowest in the youngest site. The 2000-yr-old site
where no responses to N or P were seen had trees with
leaves with the highest concentrations of N and P.
On physically different (but chemically similar) sub-
strates different elements limit growth. Raich et al.
(1996) showed that trees growing on soils derived from
lava 140 yr old (which weathers slowly to produce thin
soils) responded to both N and P. In contrast, trees
growing in soils developed over 200-yr-old unconsol-
idated ash (which was chemically similar to the lava)
responded to N but not P (Vitousek et al. 1995a).
The tendency to move from N limitation on younger
soils, especially those that are derived from unconsol-
idated substrates, to P and N limitation on older soils,
is consistent with the idea that N is accumulated from
the atmosphere during soil development, while P is
derived from the substrate. Thus N accumulates during
soil development but P, which is relatively available in
young soils, becomes progressively bound up in in-
soluble and/or physically protected forms in very old
soils (Walker and Syers 1976, Crews et al. 1995). It is
not clear whether the results from Venezuela (where N
increased litterfall and P had no significant effect) and
Jamaica (where both N and P limited growth) are in-
consistent with this; it depends on the age of the soils.
In general, in both Venezuela and Jamaica the soils are
old in that they are not derived from geologically recent
volcanic activity; however, both sites are ;500 m be-
low the current tree line, which was much lower in the
ice ages, when the sites would have been paramo. Both
sites have actively weathering rock fragments ,2 m
from the surface and these are probably a source of P.
Thus if sites are to pass from N limitation to P limi-
tation they probably have to be on deep soils that de-
velop over old, nearly level surfaces; the majority of
montane soils are on slopes where erosion rates have
probably been sufficient to ensure that there is always
Ecology
Vol. 79, No. 1
weatherable P accessible to plants. Thus the results
from the experiments carried out so far support the
general model of Walker and Syers (1976) that plant
production should be limited by N on young soils and
by P on old soils; N and P supply should more or less
equilibrate on intermediate-aged soils.
CONCLUSIONS
Results from this limited set of experiments suggest
that nutrient limitation to aboveground net primary pro-
duction may be widespread in montane tropical for-
ests—and that the particular nutrient(s) that limit(s)
production may vary among sites for explicable rea-
sons. Similar fertilization studies have been started in
tropical lowland rain forests in Panama and Kaliman-
tan. In Panama, Cavelier (1989) showed that fertiliza-
tion with N and P together had no effect on litterfall
in 4 mo after first fertilization; with hindsight we can
suggest that any effects would have been unlikely to
appear in such a short time. In Kalimantan, on sandy
soils, N increased girth increment for some trees and
N plus P together increased fine litterfall 18–24 mo
after first fertilization (Mirmanto 1996; J. Proctor per-
sonal communication). We await further results from
these studies and from fertilization experiments at two
altitudes in Puerto Rico with great interest. At present
we have insufficient information to say whether in gen-
eral montane forests differ from lowland forests in the
extent of nutrient limitation or even which nutrients
generally limit growth, despite indirect evidence that
they might differ in both respects.
Comparisons between responses to fertilization and
foliar or litterfall nutrient concentrations (or soil nu-
trients) can only be made for a few sets of data, but
obviously such relationships are potentially very valu-
able because they tell us to what extent simple(r) mea-
sures can be used to detect nutrient limitation (deter-
mined by fertilizer experiments). The best data come
from Hawai‘i where Metrosideros trees with lowest N
concentrations in leaves show the highest increase in
trunk growth after fertilization with N; an analogous
relationship was not found for P. The other experiments
have much less background information because they
are in much more species-rich forest, in places that are
much less well studied. We still need many more ex-
periments (especially in the lowlands) about the re-
sponses to individual nutrients in a range of sites—and
clearly negative as well as positive results need to be
published—before we can have confidence about the
general patterns of nutrient limitation in tropical mon-
tane and lowland forests.
LITERATURE CITED
Berendse, F. 1993. Ecosystem stability, competition, and nu-
trient cycling. Pages 409–431 in E. D. Schulze, and H. A.
Mooney, editors. Biodiversity and ecosystem function.
Springer-Verlag, Berlin, Germany.
Bloom, A. J., H. A. Mooney, O. Bjorkman, and J. Berry.
 January 1998 TROPICAL MONTANE FORESTS
1980. Materials and methods for carbon dioxide and water
exchange analysis. Plant Cell and Environment 3:371–376.
Bruijnzeel, L. A., and J. Proctor. 1993. Hydrology and bio-
geochemistry of tropical montane cloud forests: what do
we really know? Pages 25–46 in L. S. Hamilton, J. O. Juvik,
and F. N. Scatena, editors. Tropical montane cloud forests:
proceedings of an international symposium at San Juan,
Puerto Rico. East-West Center, Honolulu, Hawaii, USA.
Bruijnzeel, L. A., M. J. Waterloo, J. Proctor, A. T. Kuiters,
and B. Kotterink. 1993. Hydrological observations in
montane rain forests on Gunung Silam, Sabah, Malaysia,
with special reference to the ‘‘Massenerhebung’ effect.
Journal of Ecology 81:145–167.
Cavelier, J. C. 1989. Root biomass, production and the effect
of fertilization in two tropical rain forests. Dissertation,
Cambridge University, Cambridge, UK.
Chapin III, F. S., P. M. Vitousek, and K. Van Cleve. 1986.
The nature of nutrient limitation in plant communities.
American Naturalist 127:48–58.
Crews, T. E., K. Kitayama, J. H. Fownes, R. H. Riley, D. A.
Herbert, D. Mueller-Dombois, and P. M. Vitousek. 1995.
Changes in soil phosphorus fractions and ecosystem dy-
namics across a long chronosequence in Hawaii. Ecology
76:1407–1424.
Cuevas, E., and E. Medina. 1988. Nutrient dynamics within
Amazonian forests. II. Fine root growth, nutrient avail-
ability and leaf litter decomposition. Oecologia 76:222–
235.
Edmisten, J. 1970. Soil studies in the El Verde Rain Forest.
Pages H79–H87 in H. T. Odum, and R. F. Pigeon, editors.
A tropical rain forest. A study of irradiation and ecology
at El Verde, Puerto Rico. TID-24270 PRNC-138. United
States Atomic Energy Commission, Washington, D.C.,
USA.
Field, C. B., F. S. Chapin III, P. A. Matson, and H. A. Mooney.
1992. Responses of terrestrial ecosystems to the changing
atmosphere. Annual Review of Ecology and Systematics
23:201–35.
Gerrish, G., D. Mueller-Dombois, and K. W. Bridges. 1988.
Nutrient limitation and Metrosideros dieback in Hawai‘i.
Ecology 69:723–727.
Grubb, P. J. 1977. Control of forest growth and distribution
on wet tropical mountains: with special reference to mineral
nutrition. Annual Review of Ecology and Systematics 8:
83–107.
Grubb, P. J., I. M. Turner, and D. F. R. P. Burslem 1994.
Mineral nutrient status of coastal hill dipterocarp forest and
adinandra belukar in Singapore: analysis of soils, leaves
and litter. Journal of Tropical Ecology 10:559–577
Healey, J. R. 1989. A bioassay study of soils in the Blue
Mountains of Jamaica. Pages 273–287 in J. Proctor, editor.
Mineral nutrients in tropical forest and savanna ecosystems.
Blackwell Scientific, Oxford, UK.
Heaney, A., and J. Proctor. 1989. Chemical elements in litter
in forests on Volcan Barva, Costa Rica. Pages 255–271 in
J. Proctor, editor. Mineral nutrients in tropical forest and
savanna ecosystems. Blackwell Scientific, Oxford, UK.
Hedin, L. O., J. J. Armesto, and A. H. Johnson. 1995. Pat-
terns of nutrient loss from unpolluted, old-growth temper-
ate forests: evaluation of biogeochemical theory. Ecology
76:493–509.
Herbert, D. A., and J. H. Fownes. 1995. Phosphorus limi-
tation of forest leaf area and net primary productivity on
highly weathered tropical montane soils in Hawaii. Bio-
geochemistry 29:223–235.
Jordan, C. F., R. L. Todd, and G. Escalante. 1979. Nitrogen
conservation in a tropical rain forest. Oecologia 39:123–
128.
Kira, T., and Y. Shidei. 1967. Primary production and turn-
21
over of organic matter in different forest ecosystems of the
Western Pacific. Japanese Journal of Ecology 17:70–87.
Körner, C. 1989. The nutritional status of plants from high
altitudes. Oecologia 81:379–391.
Lyford, W. H. 1969. The ecology of an elfin forest in Puerto
Rico. 7. Soil, root and earthworm relationships. Journal of
the Arnold Arboretum 50:210–224.
Marrs, R. H., J. Proctor, A. Heaney, and M. D. Mountford.
1988. Changes in soil nitrogen—mineralization and nitri-
fication along an altitudinal transect in tropical rainforest
in Costa Rica. Journal of Ecology 76:466–482.
Matson, P. A., and P. M. Vitousek. 1987. Cross-system com-
parison of soil nitrogen transformations and nitrous oxide
flux in tropical forest ecosystems. Global Biogeochemical
Cycles 1:163–170.
Medina, E., E. Cuevas, and P. L. Weaver. 1981. Composición
foliar y transpiracion de especies leñosas de Pico del Este,
Sierra de Luquillo, Puerto Rico. Acta Cientifica Venezolana
32:159–165.
Miller, H. G., J. M. Cooper, and J. D. Miller. 1976. Effect
of nitrogen supply on nutrients in litter fall and crown
leaching in a stand of Corsican pine. Journal of Applied
Ecology 13:233–248.
Mirmanto, E. 1996. A lowland rain forest fertilization ex-
periment in central Kalimantan, Indonesia. Thesis. Uni-
versity of Stirling, Stirling, Scotland.
Nadkarni, N. M., and T. J. Matelson. 1992. Biomass and
nutrient dynamics of fine litter of terrestrially rooted ma-
terial in a neotropical montane forest, Costa Rica. Biotro-
pica 24:113–120.
Proctor, J., J. M. Anderson, and H. W. Vallack. 1983. Com-
parative studies on soils and litterfall in forests at a range
of altitudes on Gunung Mulu, Sarawak. Malaysian Forester
46:60–76.
Proctor, J., C. Phillips, G. K. Duff, A. Heaney, and F. M.
Robertson. 1989. Ecological studies on Gunung Silam, a
small ultrabasic mountain in Sabah, Malaysia. II. Some
forest processes. Journal of Ecology 77:317–331.
Raich, J. W., and K. J. Nadelhoffer. 1989. Belowground car-
bon allocation in forest ecosystems: global trends. Ecology
70:1346–1354.
Raich, J. W., R. H. Riley, and P. M. Vitousek. 1994. Use of
root ingrowth cores to assess nutrient limitations in forest
ecosystems. Canadian Journal of Forest Research 24:2135–
2138.
Raich, J. W., A. R. Russell, T. E. Crews, H. Farrington, and
P. M. Vitousek. 1996. Both nitrogen and phosphorus limit
plant production on young Hawaiian lava flows. Biogeo-
chemistry 32:1–14.
Silver, W. L. 1994. Is nutrient availability related to plant
nutrient use in humid tropical forests? Oecologia 98:336–
343.
Sollins, P., G. P. Robertson, and G. Uehara. 1988. Nutrient
mobility in variable- and permanent-charge soils. Biogeo-
chemistry 6:181–199.
Smith, J. A.C. 1991. Ion transport and the transpiration
stream. Botanica Acta 104:416–421.
Tanner, E. V. J. 1977. Four montane rain forests of Jamaica:
a quantitative characterization of the floristics, the soils and
the foliar mineral levels, and a discussion of the interre-
lations. Journal of Ecology 65:883–918.
. 1980. Studies on the biomass and productivity in a
series of montane rain forests in Jamaica. Journal of Ecol-
ogy 68:573–588.
Tanner, E. V. J., V. Kapos, and W. Franco. 1992. Nitrogen
and phosphorus fertilization effects on Venezuelan montane
forest trunk growth and litterfall. Ecology 73:78–86.
Tanner, E. V. J., V. Kapos, S. Freskos, J. Healey, and A. M.
Theobold. 1990. Nitrogen and phosphorus fertilization of

22 SPECIAL FEATURE
Jamaican montane forest trees. Journal of Tropical Ecology
6:231–238.
Uehara, G., and G. Gillman. 1981. The mineralogy, chem-
istry, and physics of tropical soils with variable charge
clays. Westview, Boulder, Colorado, USA.
Veneklaas, E. J. 1991. Litterfall and nutrient fluxes in two
montane tropical rain forests, Colombia. Journal of Trop-
ical Ecology 7:319–336.
Vitousek, P. M. 1982. Nutrient cycling and nutrient use ef-
ficiency. American Naturalist 119:553–572.
. 1984. Litterfall, nutrient cycling, and nutrient lim-
itation in tropical forests. Ecology 65:285–298.
Vitousek, P. M., G. Aplet, D. Turner, and J. J. Lockwood.
1992. The Mauna Loa environmental matrix: foliar and
soil nutrients. Oecologia 89:372–382.
Vitousek, P. M., and H. Farrington. 1997. Nutrient limitation
and soil development: experimental test of a biogeochem-
ical theory. Biogeochemistry 37:63–75.
Vitousek, P. M., G. Gerrish, D. R. Turner, L. R. Walker, and
D. Mueller-Dombois. 1995a. Litterfall and nutrient cy-
Ecology
Vol. 79, No. 1
cling in four Hawaiian montane rainforests. Journal of
Tropical Ecology 11:189–203.
Vitousek, P. M., and R. L. Sanford. 1986. Nutrient cycling
in moist tropical forest. Annual Review of Ecology and
Systematics 17:137–167.
Vitousek, P. M., D. R. Turner, and K. Kitayama. 1995b. Foliar
nutrients during long-term soil development in Hawaiian
montane rain forest. Ecology 76:712–720.
Vitousek, P. M., L. R. Walker, L. D. Whiteaker, and P. A.
Matson. 1993. Nutrient limitation to plant growth during
primary succession in Hawaii Volcanoes National Park.
Biogeochemistry 23:197–215.
Vitousek, P. M., L. R. Walker, L. D. Whiteaker, D. Mueller-
Dombois, and P. A. Matson. 1987. Biological invasion by
Myrica faya alters ecosystem development in Hawaii. Sci-
ence 238:802–804.
Walker, T. W., and J. K. Syers. 1976. The fate of phosphorus
during pedogenesis. Geoderma 15:1–19.
Whitmore, T. C., and C. P. Burnham. 1969. The altitudinal
sequence of forests and soils on granite near Kuala Lumpur.
Malayan Nature Journal 22:99–118.